Reballing ballscrews - lots of questions

Cblair suggested I post this here, and see how the thread progresses.

I'm redoing the X and Y ballscrews in a smallish VMC, 18 X 30, 25 HP, square ways, 600 IPM max. 2" diameter screws, .400" pitch. They seem to be in good shape but I have the machine apart to inspect and clean the Turcite and wanted to go over the ballscrews at the same time before I put it back together and into service. They are a lot of work to get to, I don't want to do this again anytime soon.

So I have lots of questions, this being my first time.


1. - The big one - Which is preferrable, all the balls the same size, or using spacer balls?
I spoke to Express Ballscrew Repair a bit and he strongly preferred using spacer balls, because he felt he could use more preload to get a tighter more accurate screw. The concensus of previous PM threads is that using all the same size balls increases the load capacity and that the screw will last longer with less wear. So are both of those opinions true, and which is preferrable?

2. - This machine has glass scales on all 3 axes. Does having scales change how much preload you would like to have in the screw? Since the scales determine position instead of the screw, I can see that you could maybe run less preload, possibly have less wear and not lose accuracy. If less preload is necessary, perhaps there's no downside to using all the same size balls.

3.- If you have the preload set with spacer balls, then reload the screw with all the same size balls, the same size as the larger balls previously, will the drag increase?

4.- With alternating ball size, does the size of the spacer balls matter? When replacing the balls with new, do you replace the spacer balls also?

5.- How do YOU measure ball size? I've read about elastic deformation of balls measured with a micrometer. I'm using a comparator micrometer checked against gage blocks. I have no good way to measure the force of the anvils on the ball, but my guess is it's slightly less than a starrett or B&S mic cranked down with the thimble. If the gage blocks that make the comparator mic read the same as the balls add up to .2584, what size are the balls? I can discern closer than .0001 with this mic, easily .000025, and it repeats, can't speak for it's true accuracy except to compare to gage blocks.

6.- How much ball size change is needed to make a significant preload change? If I have a screw with what seems like zero clearance, it has no measurable movement between the screw and nut and turns easily, how much larger do you go with the balls in one step to make a change?

Guess that's enough questions to start a thread. I'll have more.

Very interesting discussion, I hope we get a good number of educated replies. Some of your questions are pretty technical, I wonder if your previous source might not be the best place to get answers, but I still wait to hear what folks here have to say.

Charles

It sure seems like using spacer balls plus increasing preload* would play havoc with screw life. You would have twice the stress immediately, due to having half the number of supporting balls, and then you would heap on more stress with higher preload. I also question how how using spacer balls would result in more accuracy, because the assembly with fewer supporting balls would be more compliant (springy). You would necessarily have to increase the preload to make up the loss in stiffness. If the OEM used spacer balls, then maybe they considered it in the design, but if they did not, I'd be very wary.

*By preload, I mean load between the nut and screw through the balls, not axial stretch preload of the screw.

I'm no expert, I send all our ballscrew's out to be rebuilt.
Be ready for a headache if you decide to pull a ball nut apart yourself. Ask me how I know.....
It's not as easy as just putting in new balls etc.
All our rebuilt screws are preloaded from 6-8 INCH lbs.
They achieve this by grinding the screw, resizing balls, etc., and probably much more I'm unaware of.
If your ballscrews are not totally wasted you may be time and money ahead to just send them out to be "freshened up" for a few hundred bucks.
I send all of ours to Lead Screw Intl. , I highly recommend them.

A few more thoughts,
It would be advisable to do check backlash with an indicator.
My methods of testing a ballscrew, as learned from those that know, is to set the ballscrew on it's end.
the ballnut should have enough preload to hold it's own weight.
If it's "notchy" or rough feeling, it will be obvious it needs reconditioning.
If your's pass these tests, I don't think I would do much more than clean them up, reinstall,(possibly with new bearings?)
make sure everything is properly lubed, an run it.

I never did a large or screw but I ordered ball sets in .ooo1 increments.
Take off the transfer tube and count the balls, color one ball with a marker
and start winding in the new balls until the marked one comes out the fill the transfer
and put it on, if a tenth doesn't tighten it up then try 2 tenths.
Question 2. we only allowed a tenth or less backlash , the screw needs angular contact on one end
thrust bearings are sometimes used on the other.

Never did one but have sent them out.. Top shop IMHO

Ball Screw Repair, Refurbishment & Reverse Engineering Services - Sterling Heights, MI

Bump for anyone interested who hasn't seen it.

I was really hoping this thread would take off and get in to some good details.
Ballscrew guru's, where are you?

I reballed my x and y screws earlier this year, here is that thread http://www.practicalmachinist.com/vb/cnc-machining/reballing-some-screws-have-question-283748/. This thread contains more information than I am putting here. I am getting ready to do some more work to this mill. When I do I will test the drag torque on the nuts to see if they have changed and then start a new, organized thread on reballing a screw based on what I have learned.

Do your screws or nuts have part #s? If so call the manufacturer and get as much information as you can, especially drag torque. You need to know what drag torque to set your nut to and you have to measure it with a scale, no guessing. My info is based on my screws. 32mm diamater, 16mm pitch, 1/4" balls, 36 per side, 72 per nut, no spacer balls. Mill is 18 years old and I have had it for 10.

1) Do you currently have spacer balls? I would say stick to what you have. Or, ask the engineer when you get the drag torque specs.
2) No, set to spec.
3) Yes, by around 60%.
4) Ask the engineer. I would replace them. On size balls are 1/10th the price of oversize balls.
5) I used my tenths Starratt mic, zeroed using an on size ball. It was good to .0003 differences in ball size, which made them nearly useless.
6) 30 millionths or so for my screws.

I got my balls from Baltec at https://www.precisionballs.com/ and will do so when I do my z screw. As per their suggestion I got a few balls, 14, of various sizes plus 100 1/4" balls to use as spacers to figure out what size of ball each nut needed. Once I got some drag on the nut I then ordered sets of that size. When I installed the full set of oversize balls I found they were way too tight, turned out the balls were around .0001 to large.

The y axis on my mill has never had use on the last 6 inches of travel. When I tested drag torque I did not see any difference between the heavily used, abused, section vs the unused section. This leads me to believe the screw wears very little. Before I reballed this screw it had around .007 of backlash, perhaps more. Hard to measure since the ways are so shot.

I'm no expert, but I did rebuild the THC balls crew on my Matsuura. THC (I think it was THC) wasn't that forthcoming, they wanted me to send it in at a fairly outrageous price, but I did coax the ball size out of them and that all the balls were the same size if that worth anything to you. The spec'd size in my screw was 0.250", and that's what I put pack in it. Got em straight from Mcmaster, and even through that might not be kosher, it was still a helluva improvement over the rusted shit that came out of it.

Brent

DavidScott - I did read your thread, it was part of my inspiration to start this one. It's probably the most informative general thread about ballscrews here.

My screws don't have much ID on them. They are different styles, the X has external returns and has a Warner sticker on it and a few numbers hand scratched on it. I'll call Warner but I don't expect much free help.The Y screw has internal returns and had a Sargent Industries sticker on it. and no other markings. The only reference to Sargent I've found in Google so far is about a Superfund cleanup of the demolition site where their ballscrew plant was torn down in LA.

I took the X apart, cleaned it up and reassembled it. It has spacer balls, and has what subjectively feels like a normal amount of drag. The larger ball dia is .258x, which makes me think it was reground since the balls are .008 over normal. The screw and nut look fine, I believe I could reassemble it and put it back in service as is, but I'd like to put news balls in just to appease the gods (gremlins?) that govern mechanical devices by spending some money on it to make it 'fresh'. (Actually since the balls fail before the screws wear, I'd like to put new balls in just to restart the clock.) The spacer balls are 2 different sizes, .0005 under and .001 under (approximately). As least I'm assuming they are spacer balls, I just dumped them all out and started measuring, I didn't check to see if they alternated . (maybe they are all supposed to be the same? is .0005 enough smaller to be a spacer? .0075 over is still a pretty big oversize.)

I took the Y apart tonight. It has a seal spacer/sleeve shrunk on one end that had to be removed to unscrew it from the nut, express told me to try heating it or else machine it off. I heated it and drove it off, may have to make a new one. The balls and nut are soaking, preliminary measurement shows the balls are all about .2495 measured with a normal B&S 0-1 mic. I'll check again when they are clean. The screw looks ok under the varnish/dirt, further cleaning will show more. In the center where it was clean, it had no drag when turned by hand. So I'm wondering whether it had drag when installed and has worn slightly, or if the non-spacerball screw had less preload from new.

The X had 160 balls, the Y has 145 so far, I might get one or 2 more out of the internal returns once they are clean.

I have the ballscrews from an identical parts machine to play with. They are both external return, but still a little different from each other and different from the original X. One had the returns held in by one retainer each, the other had the returns held in by epoxy encapsulation. I took both apart, both use spacer balls. In both the larger balls are close to .250, around .251-.252.


Have any opinion on mixing and matching screws and nuts? Can you put a nut of one brand on a screw made by another manufacturer?

This discussion has been very interesting, esp about the different size balls. I have heard of this before but wasnt sure why it was done. Some of the suggestions made here have been very helpful in understanding it.

I do have a related question, when speaking of preload on the the ballnut how is that measured?

Charles

More data. The Y has 2 sizes of balls, .2497 and .24945, only .00025 difference. Before I cleaned them because they were so close I thought they might be all the same and I was just mismeasuring them. Now I wish I had been more careful on disassembly, I'll have to reload them and measure the backlash just to make sure they are alternating and not one size in one track and one size in the other.

CBlair said:

This discussion has been very interesting, esp about the different size balls. I have heard of this before but wasnt sure why it was done. Some of the suggestions made here have been very helpful in understanding it.

I do have a related question, when speaking of preload on the the ballnut how is that measured?

Charles

Click to expand...

DavidScott described that pretty well in his thread, seems his were measured by mearuring the drag while turning at a particular RPM. I believe the alternating balls are to prevent the load bearing balls from skidding, the small balls work like idlers turning the opposite direction, plus the balls don't slide against each other.

Forgive me if this was mentioned before, but what kind of VMC is this? 600ipm on all axis? Boxway/turcite on all axis (if the ways are un-even at all or if theres any sagging going on that will loosen up a tight ballscrew in a hurry)? How is the ballscrew supported, both ends? How about thrust bearing setup? Any stretch going on?

Edit: Regarding the use of glass scales, it depends on the specs of the scales and servo encoders and how the control is setup. What control is on here, and what are the specs on the servos and scales?

IMO, check the backlash between the nut and the screw with a rigid indicator setup. Rotate the screw by hand and see how much lash you can get with no load, then with load. There is more "backlash" going on while machining with the inertia and box-ways, sagging, servo response, control, etc.

RevetsP said:

Forgive me if this was mentioned before, but what kind of VMC is this? 600ipm on all axis? Boxway/turcite on all axis (if the ways are un-even at all or if theres any sagging going on that will loosen up a tight ballscrew in a hurry)?

Click to expand...

Monarch, 600 on X and Y, not sure about Z today but I'm only working on X and Y, yes.

How is the ballscrew supported, both ends? How about thrust bearing setup? Any stretch going on?

Click to expand...

Double axial bearings (MM9310 WI 2H DUH) and one radial on both ends, no stretch.

Edit: Regarding the use of glass scales, it depends on the specs of the scales and servo encoders and how the control is setup. What control is on here, and what are the specs on the servos and scales?

Click to expand...

Dynapath control, Siemens drives and servos, not sure what model ATM, 20 years old so it's not going to have huge accelerations. Heidenhain 703 scales, at least .0001 resolution I believe because Monarch specs repeatability to .0002

IMO, check the backlash between the nut and the screw with a rigid indicator setup. Rotate the screw by hand and see how much lash you can get with no load, then with load. There is more "backlash" going on while machining with the inertia and box-ways, sagging, servo response, control, etc.

Click to expand...

No backlash detectable on either with no load. It's all apart ATM so I can't load them to test.

Did you ever get your balls figured out? You said in another post that you were having problems with variations on one size. How much variation and how are you measuring them? I had one bag that I swore there was around .0002-.0003 variation. I was measuring with my tenths mic with a friction thimble and repeatability was around +/-.0001 so I didn't trust my own method of measuring enough to question it too much, especially since I didn't use them anyway.

Great thread.....I'm approaching a job like this on a little Bridgeport with a CNC retrofit. Non-profit, no money. We have experience with ball screws in general, just not the ones on machines.

In my research on this topic, the reason why spacer balls DO increase capacity is because the limiting factor is the wear between the load-bearing balls. The ball surfaces are scrubbing in opposing directions. The non-load bearing ball is able to run in the opposite direction and provides the load bearing balls with bearings against each other. Apparently they figure ball-to-ball wear limits the load more than the contact pressure with the screw/race.

But....I'm not a direct expert, just what I read. Very interesting stuff.

DavidScott said:

Did you ever get your balls figured out? You said in another post that you were having problems with variations on one size. How much variation and how are you measuring them? I had one bag that I swore there was around .0002-.0003 variation. I was measuring with my tenths mic with a friction thimble and repeatability was around +/-.0001 so I didn't trust my own method of measuring enough to question it too much, especially since I didn't use them anyway.

Click to expand...

I did get it assembled. How I first noticed the variation was while sorting the two sizes after a trial assembly and then disassembly. I have indicating micrometers we use for close tolerances in the turning centers, I was using one to speed up sorting the big balls from the small balls when I discovered I had a lot more spacer balls than load balls after measuring them for sorting.

I'm measuring the balls by interpolating between the lines of .0001" indicators so I'm sure the metrology purists will be flaming me but I have nothing better at hand. I'm comparing the indicating mic to a .0001" indicator with a large dial on a stand and to gauge blocks. It's easy to repeat my measurements and get the same reading, and I had another machinist try his hand at my methods and he quickly got the same results so I'm somewhat confident that I'm getting good comparisons. The mic and the indicator produce the same results.

Here's the comparator mic -


And here's the indicator method - I'm showing 2 balls from the same bag - you can see a .0003 difference.


I sat down to measure all the balls I had and hoped to come up with enough of the same size from all the bags I had received to put the ballscrew together without ordering any more. I was prepared to put one size in one track and another size in the other track if necessary. I discovered that one bag of balls that was labeled as smaller than I wanted and shipped to me in error was actually larger than the balls that were labeled the correct size. I measured them all and 80 of the 90 were just what I wanted, and I only needed 80, so those were what I used. Of the other 10, 1 was about .000010 smaller (I THINK) and the rest were about .000050 to .000080 smaller. These were 440 stainless, and this was the most consistent lot I received, the rest were chrome steel.

Of the others, the most of one size in any of the bags was 52, with all the varying balls being smaller by .up to .0003 in small groupings of 3 to 8 balls.

I had hoped to get balls of known size measured by the manufacturer, measure them with my methods and be able to estimate how much my measurement method varied from theirs so that in the future I'd be able to make an educated guess at how much bigger of a ball I needed to order to get the size I wanted. Unfortunately I don't have a ball on hand that I KNOW is a certain size, so I'm still guessing. I think my methods show 15 to 20 millionths smaller than actual, but that's just a guess from making about 700 measurements and guessing at which of the balls I received are closest to the labeled size.

The major lessons for me are 1)to do this myself when at all possible, because if a rebuilder just opened the bag and filled the screw, who knows how many of the balls are the correct size and how many are either overloaded or not carrying the load, and 2) order lots more balls than I need and sort them to get enough of the right size.

The variations I experienced may explain why you got unexpected reading with your mic, and why the sample balls you ordered for your first assembly didn't give you the same preload as the larger quantity did.

jbrowne said:

Great thread.....I'm approaching a job like this on a little Bridgeport with a CNC retrofit. Non-profit, no money. We have experience with ball screws in general, just not the ones on machines.

In my research on this topic, the reason why spacer balls DO increase capacity is because the limiting factor is the wear between the load-bearing balls. The ball surfaces are scrubbing in opposing directions. The non-load bearing ball is able to run in the opposite direction and provides the load bearing balls with bearings against each other. Apparently they figure ball-to-ball wear limits the load more than the contact pressure with the screw/race.

But....I'm not a direct expert, just what I read. Very interesting stuff.

Click to expand...

I believe you are correct. Needle bearings roll against each other and I think they are all the same size, so I thought ballscrew bearings could work the same way, especially since 2 spheres touch only at a point, and the balls are loaded against the screw and nut but not assembled with a load against each other.
I made an impromptu test - I had a helper counting balls while I loaded them and we lost track of how many of each size were loaded when I was about half done. So I put all load balls in until they were exhausted, then finished with smaller spacer balls. Assembled like that the screw was hard to turn and very rough feeling, with light white grease for lube. That told me enough to forget trying using all load balls.

What did Balltech have to say about the variations? .0003 difference is huge!
The variations I was getting with my mic was with one ball so can't blame the balls for that. When I have time I will set up my tenths indicator and test the balls for size again. I bought bags of 100 so I still have 28 left of what I used in my nuts.
I am pretty sure that the reason my sample balls preloaded looser was because 9 balls will create less preload than 64, assuming they are the same size.